Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
\JPs I23POLIcy RESEARCH WORKING PAPER 1623
Environmental Degradation Because of the importantrolethat children play in coilection
and the Demand for activities, the dernanc for
Children children may increase as local
environmental resources are
depleted, setting up a viJious
Searching for the Vicious Circle circle between resourcedepletion and population
growth. Analysis ofDeon Filmer
household data frorm PakistanLant Pritchett
yields some support for tiis
hypothesis, although the
effect may be small and
dependent on endogenous
local property rights.
The World BankPolicy Research DepartmentPoverty and Human Resources DivisionJuly 1996
I POI.IC RESEARCH WVORKING PAPER 1623
Summary findings
Filmer and Pritchett explore the hypothesis that - * Collection absorbs a quarter of the time of children.because of the impportant role children play in collection * Women benefit when there are older children in theactivities (firewood, water, grazinig) - the demand for household. They work 2.6 hours a week less inchildre n may increase as local environmental resources household activities for each child aged 10 to 15, and 3.2are depleted, setting tip a vicious circle betweeni resource hours less for each child over 15.depletioni and populatioll growth. * There seems to be a relationship between fertility
U'sing a large-scale houselhold data set from Pakistan, and the availability of firewood. Even after controllingwith detailed inforniation on fertility and the allocation for other determinants of fertility in reduced formof time (of womeni, childreni, and adults) to collection regressions, they show that households that live someactivities, they find that: distance from firewood have more children whereas
Collection activities absorb a substantial part of households that live where firewood is more expensivehouselhold resources. Firewood collection accounts for have fewer children.6.2 perccnt of houselhold expenditures, valued incollection time.
T-his paper- a produrct of thie Poverty and Human Resources Division, Policy Research Department- is part of a largereffort in the department to investigate the social and environmental consequences of growth-oriented policies. Copies ofrhe paper are availahle free from the World Bank, 1818 H Street NW, Washington, DC20433. Please contactSheilaFallon,roomn N8-030, telephone 202-473-8009, fax 202-522-1IS., Internet address [email protected]. July 1996.(5U pages)
Ihi, Policy Researck Working IPaper Series disseminates the findings of work in progress to encourage the exchange of ideas aboutetCel,)plient iss4tcs. An obje,ti!e of the series is to get the findings out quickly', even if the presentations are less than fullypolished. The
pipers carry tht natnes of the authors and should be used and cited accordingly. The findings, interpretations, and conclusions are theauthors'owun and sl'nuld not be attributed to the Wor/d Bank, its Executive Board of Directors, or any of its member countries.
Prodtucced by the Policy Research Dissemination Center
Environmental Degradation and the Demand for Children:Searching for the Vicious Circle
Deon FilmerWorld Bank
Lant PritchettWorld Bank
I
E-nvironmental Degradation and the Demand for Children:Searching for the Vicious Circle'
As environmental degradation and population growth have come to the fore of the
development agenda, so too have beliefs and theories that suggest that problems of poverty,
population, and the enviroment are inter-linked. In partcular, some have proposed that
environmental degradation and population growth may exacerbate one another in a vicious circle
in which greater population leads to a worsening environment and a worsening environment leads
to more rapid population growth (Dasgupta, 1995, Cleaver and Schreiber, 1995). One half of the
vicious circle may seem obvious, that greater population in a given area leads to pressures on, and
ultimately degradation of, the natural resource base: soil quality, water, grazing areas, wood
availability2 .
The explanation of the second half to the vicious circle, that greater envirommental
degradation leads to increased population growth, is less straightforward. Most conventional
We would like to thank Kenneth Chomitz, Lawrence Goulder, the seminar participants of the1nvironmenta Policy Forum at Stanford University and the Advisory Committee for the research project,
'Social and Environmental Consequences of Growth-Oriented Economic Policies' for helpful comments.We would like to gratefully acknowledge funding from the research project.
I This proposition is not at as obvious as it may seem: some elements of the natural resource base,pardcularly the soils, may be under a regime of property rights such that an increase in population density,given the right policy, leads to higher prices and increased investments such that soil qualky increases withgreater populadon density. A recent study of the Machakos district of Kenya from 1930 to 1990 showsthat a five-fold increase in population lead to no degradation (and possibly improvement) of soil quality,as firmers invested in terracing and other agricultural improvements (English, Tiffen and Mortimore,1994).
1
theories suggest that greater natural resource scarcity should lead to a lower demand for children
and slower population growth, both dtrough the direct effect of lower income in reducig demmd
and because the scarcity of resources reduces the productivity of children as household produes.
In this case the fertility response to environmental degradation mitigates, rather than exacerbas,
environmental problems3.
What is the behavioral mechanism that would produce a vicious circle? Greater
environmental degradation could lead to increased population growth if increased scarcity of
natural resource goods leads to a higher relative value of children. This is possible if children's
comparative advantage as household producers is the acquisition of natural resources for which
the family does not incur the full cost (for example, the collection of firewood, fetching of water,
or grazing of livestock on open access land). It is possible this increase in the relative value of
children from increased scarcity of available environmental goods could outweigh the income and
productivity effects from privately owned resources and lead to a higher demand for children.
This paper empirically examines the links between children and the environment using data
from the 1991 Pakistan Integrated Household Survey (PIHS)4. This survey is particularly valuable
in that it contains four types of data relevant to the vicious cycle which are rarely found together;
a) detailed information about the use of firewood and other energy sources (including their
3 Although this is not to say the fertility response will be sufficiently strong to avoid either a Malthusianlow level equilibrium or worse, an environmental crisis, in which population pushes the natural resourcebase beyond some 'breakdown" point, or especially a degradation of specific parts of environmentresources such as biodiversity. Negative (or mitigating) feedback only implies the likelihood of theseoutcomes is decreased, not increased, by individuals' fertility responses.
' A detailed description of the survey methodology for this data is in World Bank (1995) and of thisand other similar Living Standard Measurement Study surveys in Grosh and Glewwe (1995).
2
sources), b) dme allocation data for children, c) fertlity data, d) household characteristics,
particularly including total expenditures. Pakisn is chosen for study beas the existence of
this rich data set5.
The paper first briefly reviews the theoretical work on the existence of a vicious circle.
The second section examines five facts relevant to vicious circle theories with the Pakistani data
and shows that while some elements of the story hold, others are more problematic. The third
section estimates the association between various measures of firewood scarcity and the fertility
behavior of rural Pakistani women.
1) A vicious cirle: A theoretical story
There are two strands to the theoretical literature on a vicious circle between population
and environmental degradation. The first strand examines the consequences a vicious circle, if
one were to exist. The second presents household models of fertility determination that suggest
there might indeed be a vicious circle.
Nerlove (1991, 1992) examines the possible consequences of the existence of a vicious
circle, defined as a two equation dynamic system relating environmental degadation (ED) and
population (N):
' The reader should feel free to insert here the joke about the drunk looking under the streetlight forhis keys that he lost elsewhere because there was light. We admit Paistan is the country being examinedonly because Pakstan is the country for which we have household income, fertliy, and child time use dataalong with extensive data about firewood use and coliection. Future research plans to examine the sameissues in Nepal.
3
ED - g(N) , where ci( YaMNO
N = AED), where aA Y8/ED>O
This merely stat mathematically that an increase in population (N) causes the environment to
deterorate (ED) while a deerioration in the environment causes population to increase. Nerlove
woris out the implications of this simple dynamic system. As with any non-linear system without
stabilizing (and in this case with destabilizing) feedback mechanisms, the dynamic implications
are both complicated and far from pretty.
In the Nerlove papers, however, the existence of a vicious circle is only asserted as a
possibility. Nerlove's justification for this assertion is:6
For example, as forests recede up the mountain sides, parents may perceive a greater benefit ofhaving an additional child to gather firewood. More reaisfically, in a poor agricultural settinglower quality environments may be associated with a greater livestock component in totalproduction. [...] Arguably, children have a comparative advantage over adults in tending livestockin contrast to the heavier labor of planting, tilling and harvesdng crops. Thus, enviromnentaldeterioration may well enhance the marginal productivity of children, at least relative to familyproductivity.
Dasgupta (1993, 1995) and Dasgupta and Maler (1995) also develop theories with similar
reasoning: that children are devoted to that part of family income which is derived from the
exploitation of naural resources for which the primary cost is the time required to coUect the
good. Hence, as the implicit price of those goods goes up, the marginal value product of children
relative to adults rises. Parents then may have the incentive to have more children in spite of the
6Nerlove also gives a ratinlition of enromental impacts on fertility through increased mortality.However, the rationale in this case is much less clear cut as while increased mortality may plausibly leadto increased fertility, whether this leads to increased popultion depends on a greater than one for onefertility response, which is empirically debatable.
4
worsening environmental conditions, and in spite of the fact that an additional child might furter
worsen these condicons for all other families.
There are two distinctive features of the vicious cycle story versus the more conventional
theories of fertility. First, they emphasize that a significant part of household income, properly
measured, is generated by activities that use natural resources for which the cost is not fully
internalized by the household7. Second, they emphasize that children as workers are not like
adults, but are relatively specialized in these natural resource related activities.
This paper wiU not elaborate a fulUy specified model of mral household behavior which
simultaneously endogenizes all the relevant household decisions (which would have to inchlde the
allocation of household members' time across activities, the decision to purchase or collect goods,
and fertlity behavior). Our much less ambitious objective is to explore various empirical
elements of the vicious circle story. Before attempting to estimate the association between
firewood scarcity (as one indicator of the environment) and fertility, the next section examines five
factual questions relevant to the plausibility of the vicious circle versus more conventional stories.
II) The vicious circle facts
The answers to the following five questions will help determine whether the interaction
with environmental factors is likely to be important for ferdlity decisions:
* How uiportant are collected goods in household resources?
' This awkward circumlocution is the result of avoiding the familiar terms like 'private property' or"commons" as in general the pattern of use and transfer rights over particular local natural resources arevery much more complex than either of these paradigmatic cases and are endogenously determined, notgiven.
5
* is firewood collected from open access or common property sources?
* Are children relatively devoted to collection activities?
* Does the presence of children alter adult's time use?
* Do rural families alter their firewood use and collection activities in response to the
presence of children?
How important are colkcted goods relative to household resources? The collected good
about which we have the most information, and which is relevant to environmental degradation,
is firewood. Firewood is by the far the predominant fuel of households in Pakistan, especially
in rural areas. Biofuels account for 86 percent and woodfuels alone account for 54 percent of
all fuel use by households.8 Ninety percent of rural households use some firewood, with some
regional variations, less in the more developed Punjab and more in the more remote regions (table
A1.1). The primary uses of energy by households are for cooking, 78.5 percent of use, and
heating, 13.2 percent (table Al.2). Households use significant amounts of firewood on a daily
basis. Among those households that use firewood the mean (median) level of firewood use is 5.8
(5) kg per day per household in urban areas, and 7.4 (6) kg per day per household in rural areas
(table Al.4)9. These values vary somewhat across regions, with much higher consumption in
I As measured in the Pakistan Tonne of Oil Equivalent, which differs slightly from the internationalstandard, Ouerghi and Heaps (1993)
9 These figures are consistent with results other from smaller surveys cited in Ouerghi and Heaps(1993). These found for example, 6.9 kg in a survey of 197 rural households in northern Pothowar Plateauand 7.8 kg from 119 households in rain fed areas. Figures based on actual consumption are significantlyhigher than other estimates of consumption based on "needs" requirements, such as a commonly cited 2.8kg per day estimate based on cooking and heating requirements.
6
Baluhhisan, where rural household consumption is 10 kg per day versus only 5 kgs per day for
rural Punjab.
Households obtain firewood either through collection or purchase. The vast majority of
urban households exclusively purchase firewood (80 percent), so for the remainder of the paper
we will focus on rural households. Most rural households either only collect (64 percent) or
combine purchasing and collecting (11 percent), which does leave a significant fraction of rural
households (25 percent) that only purchase firewood (table A1.5). These shares vary
significantly across regions, with the highest share of collection only being in the Sindh region
(76 percent) and the lowest in Punjab region (50 percent). The degree of collection varies across
expenditure groups: the poorest 50 percent of households in the Singh region has a share of
collection of 82 percent, and that in the Punjab region has a share of collection only of 59
percentl.
As part of their study of household energy consumption in Pakistan, Ouerghi and Heaps
(1993) calculated the value of collected firewood both at market prices, and by valuing the time
that entered into the collection activity (table 1). When labor time devoted to collection activities
is valued by setting the male wage rate at 4 rs per hour, and female and child time at half that,
the value of collected firewood is quite close to its value if purchased in the market. I This should
'1 That is, the poorest 50th percentile as measured by household expenditures per person. This numberis calculated for all rural areas, and then for each province separately.
" The median reported daily male wage in the PHiS sample is 40 rs per day while the median femalewage X 25 and the median child wage is 20. Since employment opportunities for females are quite limitedand female and child labor cannot be separated in this exercise, Ouerghi and Heaps' assumption of halfthe male wage undervalues female time but is a good first approximation.
7
not be surprising as many households both purchase and collect firewood and hence are likely to
be near the margin of indifference between the two.
Table 1: Value of collected firewood in rural areas.
Number of Time Average Labor cost of collection Value of Labor cost ofcollectors required, distance (rs/year) coUected collection
person to source (at various valuaions of wood at (valuedhours per of time) market difiremly for
- - year firewood pries males,
Adult Other All labor Male, (tslyear) me andmale valued at female aMn c as
made child labor fraction ofwage valuedmeinH
l _______ wage___ ______ differently exp e sAll Pakisan .91 1.36 699 1.78 2797 1958 2099 6.2%
Punjab .95 1.28 523 1.64 2092 1491 1752 5.2%
Simdh .79 1.60 1044 1.76 4176 2778 2233 8.8%
NWFP .99 1.26 760 1.75 3040 2188 2920 5.4%
BaluchLstan .75 1.41 1091 3.79 4364 2939 2819 7.9%
Source: Adapted from Ouerghi and Heaps (1993). Based on consumption of about 6 kg per household per day. Theassuned wage rate for male time is 4 rs per hour. For female and child time the wage is assumed to be 2 rs per hour.Collected wood is valued at .80 rs/kg (as opposed to .98 rs/kg the average market price) to account for lower value ofcollected versus purchased wood.
The median total value of expenditures per household (including imputations for housing
and firewood and other non-marketed production) in rural areas in Pakistan was 31,427 rs per
year"2. This implies that the imputed value of firewood is 6.2 percent of total typical household
expenditures. This aggregate figure of course masks great variations, as in cases where firewood
is scarcer or more distant it obviously consumes a larger fraction of income. For instance in
Baluchistan, where distance to firewood is more than twice the national average, the value of
12 Since the average exchange rate was 23.8 rs/$ in 1991 this is US$1,320 per household and sinceaverage household size is about 7, the median per person expenditure in rural areas in dollars would beUS$189.
8
collected firewood is almost 8 percent of total housold expnditures. Obvudly her are also
large fluctations within thee large regions.
Is firewood coUectedfrom open access or common property sources? A crucial element
of the vicious circle hypothesis is the divergence of private and public costs of the collection
activity, that is, parents are able to push some of the cost of their children's activities onto the
local environment. Table 2 shows the reported sources from which wood was collected by nual
residents by province and then separates these out, for each geographic area, the sources of
collection for the poorest 50 percent of households. Most firewood is collected from households'
own land and from "other private' land. Overall in Pakistan, state/forest land and
common/village land account for only a quarter of firewood co11ection. This is higher in NWFP
and much higher in remote and sparsely populated Baluchistan.
This prevalence of firewood collection on private land highlights several points. First,
only a small fraction of household firewood collection consists of felling trees. While purchased
firewood tends to be large branches and stems, collected firewood is more likely to be small
twigs, shrubs, leaves and roots (Ouerghi and Heaps, 1993). This composition of the types of
firewood and the prevalence of collection of firewood from 'own land" reveals the multiple uses
of individually owned trees, and trees and shnrbs that may be present on the land for other reasons
(e.g. erosion control, shade provision, boundary demarcation) have the secondary benefit of
providing firewood.
9
Table 2: Source of collected wood among rural households which collect wood, by province.
__________ j ~~~~~~~~~~~~~~~~~~~(3)4(4)
OwnLad Other Stae Waste Common / Oder Sub-Da:private forest land village land non-private
__ __ ~~~~~~land_______
All mral households
Pakistan 26.0 34.1 12.3 O.S 13.3 13.7 39.8
Punjab 35.8 33.7 7.7 0.8 8.4 13.8 30.7
Sindh 22.7 46.5 17.9 0.4 9.6 13.5 41.4
NWFP 35.9 19.7 15.3 0.0 18.2 10.8 44.3
BaluchLMs 3.3 6.6 10.9 0.0 56.0 23.1 90.0
Households in pB b 50 percntie of ex es per person
Pakistin 21.2 37.5 12.1 0.5 13.6 15.2 41.4
26.6 41.8 7.6 0.9 9.1 14.0 31.6
Sindh 9.8 45.3 18.8 0.4 13.3 12.4 44.9
NWVFP 31.9 24.1 14.7 0.0 14.7 14.7 44.1
Baluchistan 2.0 3.9 15.7 0.0 41.2 37.3 94.2
Source: PIHS
Second, a significant fraction of firewood collected off of "other private land" suggests
that, at least in many areas of Pakistan, the value of firewood has not reached the point where
exclusion or charges for collection activity, even on private land, are profitable. This brings us
back to an issue discussed in the opening section: soil, forests, water, or grazing land are not
intrinsically private property (at one extreme) or open access (at the other). The elements of
ownership of resources (use, transfer rights and residual claims) depend not only on background
social and legal conventions but also on the ratio of the value of the commodity under one regime
or another. Even with constant prices for enforcement, as a resource becomes more valuable
10
there may be an endogenous switch from a regime of open access to more limited access (for
example, restricted common property or purely private property).
One important determining factor of this regime change will be population density itself.
All else equal, as population density (and the density of economic activity) rises, the value of land
rises which increases the likelihood of a switch to private property. As discussed above, however,
increasing scarcity of private property natural resources can be expected to have a very different
impact on fertility decisions than open access natural resources: the household cannot escape the
full natural resource consequences of its fertility decisions13.
This combination of differential household behavior with respect to common versus private
property treatment of natural resources, and the endogeneity of the socially enforced property
rights over certain resources will make for interesting dynamic behavior of a fully articulated
system. It is possible that at stages of low population density initial increases in population
density will lead to greater demand for children. This greater population density will in turn
accelerate environmental degradation and raise the scarcity value of the open access resource. As
long as the resource remains open access this will (or empirically may) in turn increase the
demand for children. After a certain point, the increased population density may change the value
of the resource such that it switches from open access to more limited access, such as private
property. Once a natural resource is privatized, the negative externality to child bearing
disappears and is internalized by the household. From that point on, households would respond
to furter deterioration in availability by reducing their demand for children (unless children have
13 The differing views on the impact of populaion density on agricultural institutions and practices arewell described in Turner, Hayden, and Kates (1993).
11
a comparative advantage over adults in the market supply from private prope which is
unlikely).
Recent studies (Ouerghi 1993, Leach 1993) on tie market for firewood in Paktn
argue that something very much like the switch to a private property regime may have already
occurred or is about to occur in some, and perhaps most, regions of Pakistan. There is now a
large market for firewood and most of the new tree plantings are on large plantation style lots.
These plantation markets supply firewood indirectly from twigs and branches cut (as the main
market for plantation trees is construction poles or other uses). Figure 1 suggests something of
the time path of the market for firewood in Pakistan with an ongoing, and perhaps well advanced
in many regions, evolution from plentiful open access forest resources to scarce, market provided
firewood.
12
F1%1From foet ngemen to tree planting - otosdictated bypoutonadeari
Forst or tree cover through time Vberable, open.> e~~~~ccess KreeorCe.
1aIrnastbiaa ean speedor slowa shift fro 3 to S.
Piaty of forest\few people so littleor no coafilet overthe resource.
- Commrnity managementsystems arc found whee thereis same resource scarcity,but wbere the resource is stillseen as defensible externaly Once there are too mayn peopleand ausmgeable internly, or too many interests for the aim mnch open-access laudusing local authority of the commrunal resource. coats becomes &rm-land ifstructures, f fmanagement tenure Is ofhed.
wil begin to outweigh bneftts. Trees are plated and aRukrs for and active local management wll noew resource ceated.owneenhip and use grdually cease. Reserves. If any,are found only near wiUl have to be protectedthe the boundary; by the state.or relate only to a Rapid deforestation will then takefew high-valued phcegathered products Resources may remain in
balance for many years.
Increase in numbers of people through time
Stage I Stage 2 | Stsge 3 Stage 4 Stage 5
Source: Chart devised by Gil Shepberd, ODI, autumn 1992/spring 1993. in OnerghL 1993
Are children relatvely devoted to collection activities? In the PIHS data there are two
types of time allocation data. All individuals aged 10 and over were asked about their "labor'
time devoted to a variety of tasks, either wage labor or the production of (potentially) marketable
commodities. In addition, all females (but only females) aged 10 and over were asked about their
participation in thirteen household production tasks.
13
Tabb 3: Allocato of tmin housold activiies in hours pr week (ad by fracton of total m) m slcted bas forwomen ad giuls, by urbbAn-na resWosuc
Urban Girls Urban Rural Girls Rrl Women Ae: 10 Ae: 25Women
Fetcing waor 0.3 0.4 1.6 1.8 1.5 2.7(1.9%) (1.3%) (7.7%) (5.1%) (11.4%) (4.4%)
FirwOod 0.0 0.1 1.2 1.3 1.5 1.4collction (0.2%) (0.2%) (5.9%) (3.7%) (11.2%) (2.4%)
Annal care, 0.2 0.6 1.9 3.8 1.6 3.9fodder (1.6%) (1.9%) (9.0%) (10.3%) (12.1%) (8.2%)
Dungcake 0.1 0.2 0.5 1.1 0.4 1.1prepaation (0.4%) (0.6%) (2.5%) (3.0%) (3.3%) (2.2%)
Sub-total: 0.6 1.3 5.2 8.1 5.0 9.1Colaecon tks (4.1%) (3.9%) (25.1%) (22.2%) (38.1%) (17.2%)
Meals to wores 0.0 0.0 0.2 0.5 0.2 0.4(0.0%) (0.0%) (0.9%) (1.4%) (1.4%) (0.8%)
Grindw flour 0.0 0.1 0.1 0.4 0.0 0.6(0.2%) (0.2%) (0.5%) (1.0%) (0.2%) (0.8%)
Going to market 0.2 1.0 0.1 0.4 0.1 0.3(1.3%) (3.2%) (0.5%) (1.1%) (0.8%) (0.6%)
Stitchin 0.6 1.1 0.7 0.9 0.2 1.4(4.1%) (3.3%) (3.6%) (2.6%) (1.4%) (3.0%)
Milk, ghee 0.0 0.1 0.2 1.3 0.0 1.0PIearadon (0. I %) (0.4%) (1.0%) (3.6%) (0.2%) (2.2%)
Sub total: Other 0.9 2.3 1.3 3.5 0.5 3.6tasks (5.6%) (7.2%) (6.4%) (9.7%) (4.1%) (7.4%)
Cooking 6.9 14.2 7.3 12.3 2.6 15.9(45.3%) (44.4%) (35.1%) (33.8%) (20.2%) (33-6%)
Caesning 5.9 8.5 5.8 7.4 3.8 10.4(38.6%) (26.4%) (28.1%) (20.2%) (29.1%) (22.0%)
Cbild care 1.0 5.8 1.1 5.1 1.1 9.9(6.3%) (18.8%) (5.3%) (14.1%) (8.5%) (19.8%)
Total 15.3 32.0 20.7 36.5 13.0 48.9(100%) (100%) (100%) (100%) (100%) (100%)
Notes: Figures in paretes amre percent of ime devoted to activity. 'Girls" in this table refers to unmauried femalehousehold members less than 18 years old.
Source: PIHS
14
Table 3 shows the allocation of time to each of the thirteen household tasks for women and
for girls'4 . On average, women in Pakistan spend 35 hours per week in household activities,
while girls spend 18 hours per week on these tasks. The three most time consuming tasks for
women are cooking (38 percent), cleaning (23 percent) and child care (16 percent) which together
account for 77 percent of time devoted to household tasks.
We classify fetching water, collecting firewood, collecting fodder for animals, and dung
cake preparation as those 'collection" activities which involve natural resources. Collection
activities account for 14 percent of time for women on average and 17 percent for girls. Not
surprisingly, these activities are much more important both absolutely and as a fraction of time
for rural women. Urban women and girls spend only 1.3 and 0.6 hours per week on collection
activities, compared to 8.1 hours and 5.2 hours per week for rural women and girls. Rural
women work 4 hours more per week on household tasks, hence the difference in collection
activities is by itself greater than the total amount of labor time difference.
What types of work do children do? As suggested as part of the vicious circle hypotheses,
the percentage of time allocated to collection activities is much higher at young ages. The final
two columns of table 3 time allocation of rural females at ages 10 and 25. Whereas 10 year olds
devote 38 percent of their time to collection activities which falls to 17 percent by age 25. Figure
2 shows the steady decline in this percentage of time in collection activities by age.
4 In this discussion, "girls' refers to unmarried females less than 18 years old.
15
Figure 2
Share oftime in household activities devoted to "collection activities"Rural females apd 10 to 25
40
30-
1 5 --- - - - - - - - - - - - - - - - - - - - -
10
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Age
This relatively high frction of child tme devoted to collection activities is consistent with
a number of studies of time allocation. Table 4 shows the results of four other studies of time
allocation in Asian countries for both the total time devoted to household activities and the fraction
of that time devoted to collection activity (appendix 2 reports more detail of these other studies).
16
Tabe 4: Afocadon of time by children to collection activities from otber studies
Country Total tume to HH acdvides Fracuon of time to collection advities(hours per week) (percent)
Woman Child Woman Child
This study Average 36.5 20.7 22.2 25. 1
This Study 75th percentile 52.5 29.5 34.5 41.9
Palstan (Alderman andCbishti) 32.0 16.4
Javanese village 32.8 20.9 (E) 3.5 22.9 (F)
Nepalese village (A) 44.3 33.1 (F) 36.1 56.3 (F)
Nepalese villages (B) 45.6 20.5 (all) 26.5 68 (all)
Nepalese villages (C) 56.0 13.5 (all) 38.9 68.6 (all)
Source and notes: Pakistan. based on Alderman and Chishti (1991) table 2 based on time allocation for all rural females over10, with same classification of collection activities as our study. Javanese village and Nepalese village (A) are based on Nag,White and Peet (1980) table 10.2 and 10.4, based on femde time use. Children are ages 6 to 14. Collection activities arefirewood coUection and animal care. Nepalese Villages (B) is based on Acharya ad Bennett (1981) based on five hill andone tarai village, coildren are 6 to 15, collection activities are fuelwood collection, water collection. leaf fodder colection.Nepalese villages (C) is based on Kumar and Hotchkiss (1988) based on three hill districts, children are 6-15, collectionactivities are collection of fuelwood, water, grass, and leaf fodder and the assunption that half of grazing is done bychildren. Nepalese B and C are based on timne use of male and female children. Time use per week based on six day workweek.
Does the presence of children alter adult's time use? Although the share of time devoted
to collection activities is decreasing with age, we now turn to how much total time is spent on
collection activities by children. How important is collection done by children as a fraction of
total household time spent on various household tasks? Table 3 showed that the share of time
devoted to collection activities falls from 38 percent to 17 percent between the ages of 10 and 25.
Figure 3 shows the hours of work in household tasks per week by age for females from age 10
(the lowest age with this information in the survey) to age 65. This rises sharply from 10 hours
per week at age ten to 24 hours per week by age fifteen, to 44 hours per week by age twenty-five.
Total time devoted to household activities rises sufficiently with age that the absolute amount of
17
time in collection activities rises substanially, from 5.0 hours for a ten year old to 9.1 hours for
a women aged 25.
Figre 3
Hours per week spent by females in household activities, by age
so - - - - - - - - - - - - - -- - - - - - - - - - - - - - - - -
20 - - co - - - -o - - - - - - - - - - ,, ,, - - - - -5
50
Age
In table 5 we shift from the individual level, how much work each person does, to the
household level. How much total time is spent by all members of the household, what is its
allocation across tasks, and how much of each task is done by children? The total female time
per household in household tasks in rural areas is 78 hours a week. Since from figure 3 the
average time per wornan peaks at around 50 hours per week (at ages 30 to 35) it is obvious that
major time contributions are made by other household members. Of the total time devoted to
household tasks in mr-al areas, 18 hours a week, or 2.5 hours per day per household, are devoted
18
to collecdon acdvities, although this total varies substantially within Pakistan. Only about 17
percent of female time devoted to household tasks is contributed by children.
Table 5: Total female time devoted to household tasks in Pakistan in rural areas, by type of activity, and the fractionwhich is done by (female) children.
Total female household Share of total time whuch is performed byhours devoted to task devoted to task chidren
Fetching water 4.4 5.6% 20.3%
Firewood Collection 3.2 4.1% 19.5%
Animal care, fodder 7.9 10.1% 13.7%
Dung Cake Preparation 2.3 2.9% 13.9%
S-t(tal cnilectign 17 a 22 -8 a _ __E
Meals to Workers 1 1.3% 13.4%
Grinding Flour 0.7 0.9% 10.9%
Going to Market 0.8 1.0% 7.5%
Stitching 2.2 2.8% 20.3%
Milk, Ghee preparation 2.4 3.1% 5.2%
= Sub-total other _ 7,1 9.1 % _,______ _
Cooking 26.6 34.1% 17.2%
Cleaning 16.9 21.6% 21.6%
Child Care 9.7 12.4% 7.1%
Total 78.1 100.0% 16.6%
N I2254 2254 2254
Note: Fraction of time in task which is performed by children is only for households in which the task is performned.
Source: PIHS
Some activities, particularly fetching water (20.3 percent), collecting firewood (19.5
percent), and cleaning (21.6 percent) appear to be "child" tasks. Others, such as animal fodder
collection, are proportionately less performed by female children. Perhaps surprisingly, given
other work which suggests a large degree of substitutability of mother's and older daughters' time
19
in caring for young children, child care appears not to be a child activity. Children do contribute
a significant fraction of all hours spent in household tasks. The impact of this on mother's time
and total hours by age is examined below.
Bringing the various pieces of information together, we find that table 1 reports that
roughly 60 percent of the time devoted to firewood collection comes from women and children,
while table 5 suggests that only 20 percent of female time to firewood collection comes from
children. The composition of members of trips to collect firewood in table 6 suggests that
children account for about 15 percent of the people who go to collect firewood. Depending on
the region, there are nearly equal numbers of adult males and females. The number of children
varies from only 9 percent in Punjab and Baluchistan to 25 percent for the poorer households in
Sindh and NWFP. Together the evidence suggests that an important, but not overwhelming part
of all firewood collection is done by children.
20
Table 6: Composition of most recent trip for wood collection, in rurl areas among houeholds dmt ue firwood
_____________ Percent which were:
Number of Quliren (male servants andpeople Adult Males Adult Females and female) odhers
All Pakistan 2.24 47.41 34.85 14.23 3.51
Punjab 2.16 50.54 37.09 9.83 2.54
Sindh 2.31 42.76 33.18 20.54 3.52
NWFP 2.33 49.63 28.34 16.50 5.53
Baluchistan 2.13 43.89 41.54 8.69 5.88
Poorest 50th household per person expenditure percentil _
All Pakistan 2.23 44.38 37.60 16.57 1.45
Punjab 2.26 47.03 40.36 12.01 0.60
Sindh 2.22 36.26 36.33 25.16 2.2S
NWFP 2.35 45.26 32.55 17.97 4.22
Baluchistan 2.29 52.65 37.76 8.49 1.10
Source: PIHS
Do rural families alter their firewood use and collection activities in response to the
presence of children? Table 7 presents the association of household composition and the number
of hours devoted by mothers to various tasks. 5 The fact that time use cannot be negative implies
the impact of children has a particular non-linear structure, that is, time in each task could be
higher or lower either because it equaled zero and now positive or because it was positive and
15 It should perhaps be emphasized here that these estimates-and those subsequent equations of therelationship between household composition and time devoted to tasks-do not correspond to the experinentof exogenously adding an individual of a particular age to the household. Such an interpretation isinconsistent with the endogenous demand for children (taken up in subsequent sections) being determinedby their value from carrying out household tasks.
21
now has changed."6 Because these estimates are a combination of changes at the limit (from zero
to positive hours), and changes above the limit, the proportion of the average total change due to
the response above the limit is also reported in the table (as derived in McDonald and Moffitt,
1980).
These results indicate that the presence of children under 10 is associated with increases
in a rural mother's time in household activities, whereas each child over ten is associated with less
time in household activities. The point estimates from table 7 indicate that by age 16 a female
child has a been a net contributor to hours in household activities, with the positive contributions
of labor from 10 to 15 offsetting the additions to mother's time from years 0 to 10.17
Table 7 also presents estimates of the relationship between household composition
variables and total time devoted female time to each of several activities. These results suggest
that an additional child in the household between 10 and 15 is associated with an additional 5
hours per week devoted to all household activities by all members of the household. But females
10 to 15 years old devote on average 17 hours to household activities. The net contribution to the
household is 12 hours on average. Mother's time is reduced only 2.6 hours. This difference is
16 Tobit estimation is used because a significant number of women devote zero hours to some of thetasks (i.e. the dependent variable is censored at zero). If the underlying (non-censored) model isH*=Xp+e, then the change in observed (censored) H due to a change in one of the Xs, 6H1 bXi, is equalto PiF(z), where fPi is the Tobit parameter estimate, F() is the standard normal cumulative distributionfunction, and z is equal to Xp/a where a is the standard deviation of e. McDonald and Moffitt (1980)recommend evaluating this at the mean of the Xs. These are the average derivatives implied by tobitestimation of the relationship between mothers time devoted to various household tasks and householdcomposition in rural areas
" With discounting, this does not necessarily make a child a positive payoff, even in strictly timeterms. Assuming a discount rate of 3 percent, (which is low for rural households in Pakistan) a femalechild staying until age 18 would pay off in mother's time terms.
22
Table 7: Mean change in expected hours per week per woman in household activities, rural areas
Mothers | Sum for all household females over 10 years old
Dependent Total Child care Cooking Collec- Toua Child care Cooking Collectionvariable in hours and tion and activitiesper week cleaning activities _ cleaning
Household composition variables: Number of household members
above 15 -3.219 -0.701 -2.103 -0.412 5.973 -0.037 2.738 1.139(<0.001)* (<0.001)* (<0.001)* (0.001)* (<0.001)* (0.731) (<0.001)* (<0.001)*
between 10 and -2.611 -0.392 -1.652 -0.463 5.249 -0.325 4.165 0.91215 (<0.001)* (0.023)* (<0.001)# (0.059) (<0.001)* (0.161) ('<0.001)* (0.020)*
between 5 and 10 0.974 0.646 0.147 0.122 5.146 1.325 1.707 1.432(0.038)* (<0.001)* (0.635) (0.533) (<0.001)* (<0.001)* (<0.001)* (<O.001)*
between 2 and 5 2.153 1.309 0.577 0.277 5.987 3.080 1.875 0.567(0.001)* (<0.001)- (0.181) (0.306) (<0.001)* (<0.001)* (<0.001)* (0.203)
under 2 1.001 1.788 -0.504 -0.271 7.490 5.031 1.987 0.437(0.179) (<0.001)* (0.311) (0.390) (<0.001)* (<0.001)* (0.003)* (0.409)
Other vanables:
HH expenditure 3.640 -0.606 3.586 -0.266 3.321 -1.040 4.699 -1.663per person (<0.001)* (0.030)* (<0.001)* (0.495) (0.019)* (0.009)* (<0.001)* (0.011)*(natural log) _ __
Age 0.635 0.193 0.161 0.454(0.020)* (0.042)* (0.374) (<0.001)* _ _ _ _
Age squared -0.016 -0.005 -0.007 -0.006(<0.001)* (<0.001)* (<0.001)* (<0.001)* _ _____
Female head 1.890 0.300 3.586 -0.430(0.639) (0.800) (0.175) (0.800) _ _ _ _ __
Spouse present 6.504 -0.318 4.386 1.475(0.636) (0.948) (0.641) (0.788) _ _ _ _
Constant 23.843 4.567 8.779 -0.583 13.040 1.877 -5.083 13.188________________ (0.127) (0.397) (0.409) (0.926) (0.145) (0.450) (0.334) (0.001)*
Summary stausucs:
Share of change
rdesponsthe a 81.2% 36.4% 69.1% 48.9% 89.2% 43.3% 86.6% 52.6%
zero
Mean hours 42.6 6.1 22.8 9.7 78.5 9.7 43.5 17.6
N 2101 2101 2101 2101 2254 2254 2254 2254
Notes: P-values of tobit parameter estimates are reported in parenthesis, (*) indicates significantly different from zero at the 95 %level
Source: PIHS
23
accounted by two factors. First, child labor is not likely to be as productive and so will not
reduce adult's time one for one. Second other females in the household will also reduce their
time.
Children and collection activities. The general information on time use does not reveal
whether children actually alter the decisions of the household about the relative value of collection
activities. In the PIHS data, however, we can examine whether mrual households with children
of various ages are more or less likely to collect versus purchase firewood. Table 8 shows the
estimated effect of a change in a variable on the (1) probability of collecting only (versus
collection and purchase or purchase only) and the probability of collecting only or collecting and
purchasing (versus purchase only) and (2) the expected quantity of firewood consumed.'8 The
patterns found here are somewhat puzzling. Each additional non-child in the household is
associated with an increase in the probability of collection only, whereas each child aged 11 to
15 is associated with a decrease in this probability, each child aged 5 to 10 is associated with an
increase in this probability (insignificantly), while a very young child (under 5) is associated with
a decrease in this probability. If children from 5 to 15 were an important asset to the household
in their ability to collect firewood, one might expect that the probability of collection only would
increase with their presence. It appears that while a large household per se is more likely to only
collect, the presence of children does not increase the likelihood of collection.
18 If the underlying continuous model for collect only is C*=Xp+e, then the change in observed(binary 0/1) C due to a change in one of the Xs, 6C/ MXj, is equal to ,if(z), where pi is the probitparameter estimate, f( is the standard normal probability density function, and z is equal to Xp. In thereported results, Xp is evaluated at the means of the Xs. See foomote 16 for the derivation of thefirewood quantity estimates.
24
Table 8: Mean change in probability of firewood collecdon and, mean change in expected quantity of firewood used, ruralhouseholds.
Dependent variable: Collect only (0/1) Collect only or collect Consumption of tirewoodand purchase (011) (kg per person)
DHH expenditures per person (natural log) -0.055 -0.027 1.932(0.003)- (0.117) (<0.001)*
Household size -0.101__ _ _ __ _ _ _ __ _ _ _ __ _ _ __ _ _ _ __ _ _ _ __ _ _ ___ _ I(<0.001)*
Household size squared 0.002__________________ _ |(<0.001)*
Number of household members over 15 0.011 0.010I____________________________________ (0.032)* (0.035)*
Numnber of household members 10 to 15 -0.027 -0.014 0.003(0.013)* (0.179) (0.882)
Number of household members 5 to 10 0.009 0.005 -0.014l ____________________________________ (0.318) (0.586) (0.362)
Number of household members under 5 -0.025 -0.020 -0.009._________________ _ (0.002)* (0.008)* (0.557)
Sindh (t) 0.223 0.183 0.422(<0.001)* (<0.001)* (<0.001)*
Baluchistan (t) 0.161 0.132 0.883l __________________________________ (<0.001)* (0.001)* (<0.001)*
NWFP (t) 0.054 0.061 0.362l ______________________________ _ |(0.078) (0.032)* (<0.001)*
Summary statistics
N 2 394 2394 2394
Mean of dependent variable 0.577 0.674 0.985
Share of change due to the response l 1 65.3%above zero
Notes: (t) indicates variables for which parameter estimates reported in collect only model are for change in the variable from 0 to1. P-values of probit or tobit parameter estimates are reported in parenthesis, (*) indicates significantly different from zero at the95% level.
Source: PIHS
Similar results hold for the total use of firewood. If the presence of children lowered the
effective cost of firewood to the household, one might expect to see that households with a greater
25
child concentration would consume relatively more firewood than more adult intensive
households. The results in the third column of table 8 show that while additional household
members are associated with lower per person consumption of firewood (perhaps because of
economy of scale effects) children do not appear to affect the consumption decision differently
than adults.
l) Firewood and fertility
This section estimates directly the association between fertility and firewood scarcity; this
is done in two steps. The first step is to identify reasonable proxies for firewood scarcity. In the
second step, these proxies are introduced into a reduced form fertility regression.
A) Measures of firewood scarcity
At this stage of the research the only proxies for environmental scarcity or degradation
used are those related to firewood and its availability. Given the prominence of deforestation in
discussions of the environmental consequences of population this is a useful first pass. The PIHS
data have several measures of the scarcity of firewood, some at the household level and some at
the cluster sampling unit level."9
Households were asked three questions relating to firewood use: first, whether there was
a problem with supply; second, the average time devoted to collect firewood; and third, the
average distance to the place where firewood was collected. The answers to these household level
questions are averaged over all observations in each cluster to obtain cluster specific means. At
19 The cluster sampling unit is the smallest unit of the census.
26
the level of the cluster, price data on firewood and dung cakes were also collected directly from
a single respondent.
Table 9 reports the averages of these cluster level variables. The proportion reporting
having problems because of shortage of firewood supply is only 8 percent. This could be because
the question went beyond the ability to collect firewood and included market availability: if the
firewood market is effective then local environmental scarcity will not be reflected in an absence
of local supply on the market.
Table 9: Sumnmary statistics on cluster level measures of firewood scarcity.
Variable: Mean Standard Minimum Maximum Number ofDeviation women
Price wood in tde market (or mandi) rs/kg 0.93 0.25 0.13 1.74 2223
Problems obtaining wood because of supply(cluster mean) 0.08 0.12 0.00 0.75 3157
Hours for two way trip time to collect wood(cluster mean) 1.91 1.02 0.50 6.33 3157
Kilometers to source of wood, round trip(cluster mean) 2.91 3.53 0.00 21.50 3038
Price of dung cakes (rs/100 cakes) 11.44 5.02 3.00 25.00 483
Density of cluster (acres per HH) 6.72 9.69 0.06 53.46 2869
Source: PIHS
The average time of a two way trip to collect firewood is almost two hours, ranging from
a minimum of half an hour to a maximum of over 6 hours per trip. Kilometers to collect wood,
average almost 3 lam, and vary from 0 1am to 21.5 kmn". The price of firewood varies
I Reassuringly if one regresses the minutes per two way trip on the kilometers to source one finds areasonable relationship, i.e. that each kilometer adds about 23 minutes to the trip.
27
considerably (the average price is almost 1 rupee per kg and varies from 0.13 to 1.74) although
not as much proportionately as the distances or flmes across clusters21.
The correlations among the various cluster level measures of firewood scarcity are reported
in table 10. The pattern of correlations is reasonable. The price of wood is positively associated
with the two-way collection trip time and the price of dung cakes, a close substitute, but the
correlation between the price of wood and kilometers to the source of wood is not significantly
different from zero. Higher prices are associated with less reported problems with supply, which
again may indicate that at a higher price the market provides a supply of wood to areas with little
local supply.
Table 10: Correlations amongst various cluster level measures of firewood scarcity, rural women
Price of wood Problems Hours for two Kilometers to Price of Defhsity ofwith supply way trip source (mean) dung cakes cluster(mean) (mean)
Price of wood 1(2223)_ _
Problems with -0.1852* 1supply (mean) (2223) (3157)
Hours for two 0.2204* .0.0669* 1way trip (mean) (2049) (2748) (2748)
Kilogters to 40.0055 0. 1627* 0.5336* 1source (mean) M - (3038) (2748) (3038)
Price of dung 0.4603* 0.3761* 0.0807* -0 608* 1cakes (467) (483) (400) '!l2) (483)
Density of -0.0345 .0.0959* 0.0809* .0.0086 0.0058 1cluster (2027) (2869) (2477) (2753) (456) (2869)
Notes: (*) indicate that the correlation coefficient is significantJ1 differebt from zero at the 95% level. The number ofobservations included in each pairwise correlation is in parenthesis.
Source: PIHS
21 This is partly because the cluster data on prices were collected from a single source while the clusteraverages depend on the variability of the households surveyed.
28
B) Children ever born. birth in the past five years and firewood scarciy
The regression results presented in table 11 and 12 explore the partial correlation of
fertility with firewood scarcity. The various firewood scarcity variables are added to basic reduced
form regressions for birth in the past five years and children ever born of ever married women
(results for the basic specification, are reported in table A4. 1).
The most robust finding in table 11 is that households living in clusters with higher prices
of firewood have fewer children, and have a lower probability of having had a child in the past
five years. This result holds true when various proxies for the "remoteness' of the cluster are
added (such as distance to main market, or district capital, or railroad) to capture the possibility
that high prices are simply due to transport costs. It is also robust if the other measures of cluster
firewood scarcity are added, either individually or simultaneously. Moreover, the negative impact
of the price of wood remains, but it is not always significantly different from zero, when various
combinations of other prices are included in the regressions.22 Although an increase in the price
of firewood was thought to reduc the relative price of children by increasing their marginal value
product relative to adult wages this finding highlights the difficulties in interpreting reduced form
regressions, as without a fully specified model of household behavior one cannot identify the
causal mechanism behind any particular partial correlation.
In general, the probability of birth in the past five years is more responsive to measures
of scarcity than cumulative children ever born. For example, in the probit regressions for a birth
n When the prices of fuel-goods (kerosene and diesel) are included in the CEB [B5Y equation the p-value is 0.253 [0.001], when the prices of food-goods (basmati rice, low quality rice, chana, sugar, andonions) are included the p-value is 0.009 (0.081], with the prices of other food-goods and other non-foodgoods (soap and cigarettes) the p-value is 0.129 (0.3741, when all of these are included the p-value is 0.613[0.348]. in these specifications the estimate of the coefficient varies from -0.74 [-0.171 to -0.23 [-.05].
29
in the past five years, four of the five cluster level variables are significantly different from zero.
Again, the price of wood is strongly negatively associated with a birth. However, other
household and cluster variables are consistent with a vicious circle interpretation. Households
with greater problems with wood supply (either as measured by a household or a cluster level
variable), or households who live in clusters that are further away from a wood source, have a
higher probability of having had a birth in the past 5 years.
The vicious circle hypothesis rests on the notion that the costs of environmental
degradation are not borne by the household, i.e. that the land from which firewood is gathered
under open access like conditions. As mentioned earlier, ownership rights are likely to be the
result of an endogenous process and will most likely vary from location to location (see figure 1).
In the case under study, we can distinguish across regions within Pakistan. Table 12 reports the
same regressions as table 11 region by region.
For most of the results, the fmdings in the regionally disaggregated analysis are just
consistent with those in the national sample. However, the results for the Sindh region are
strongly consistent with the vicious circle hypothesis. Except for two (the price of wood which
is significantly negatively related to the probability of birth, and density of cluster which is
insignificantly positively related to the probability of birth), all the scarcity measures have the
expected sign and of these eight, six are significantly different from zero. Using the stages
described in figure 1 this would suggest that the Sindh region was in stage 1 or stage 2, i.e. a
region in which forest land is relatively abundant. In their description of usage patterns, Ouerghi
and Heaps (1993, p. 4-9) claim that high firewood use in the Sindh region is consistent with the
abundance of firewood rich land (i.e. riverain forests) relative to other regions, as well as a
30
relative lack of alternative fuels. They also claim that although the NWFP region has the largest
forest cover, demand for firewood may be higher due to its ngorous winter. This higher demand
could hasten the move from stage 2 to stages 3 and 4 in figure 1, i.e. contribute to a change in
ownership rights and hence we would not observe the vicious circle.
A claim that these results confirm the existence of a vicious circle between environmental
degradation and the demand for children is clearly far too strong. The results, however, are
supportive of the notion that there is a stage in the relationship between environmental
degradation, fertility, and land ownership rights in which children are in relatively high demand23.
2 These supportive, but tentative, empirical results are consistent with Cleaver and Schreiber (1994)who find that the rate of deforestation is positively related to the total fertility rate in a pure cross sectionof thirty eight African countries, but that the estimated deforestation effect on fertility was small andinsignificant in a pooled time series cross-section estimation when cultivable land area was also included.
31
Table 11: Estim of the change in the probability of a birth in response to a change in a measure of firwood scarcity in probit regesionsof a Birth in the Past S Years on the sample of, for rural, ever married women, age 19 and over and the coefficieota on mares of firevoodscarcity when entered singly into OLS regressions of Children H3Mr Bomn for rural ever-nmaried women 14 and over.
____________________ Birth in the Past 5 Years (0/1) Children Ever Born
Poorest 50th Poorest 50thExpenditure per Adult Expenditore per Adult
Variable aded All nrual Pakistan Percentile All rural Pakistan Percentile
I Mean problems with 0.319 0.203 0.621 0.715supply (cluster level) (<0.001)* (0.017)* (0.123) (0.165)_________________ 2908 16311 3184 1735
2 Mean kilometers to I 0.011 0.011 0.021 40.001source (cluster level) (<0.001)* (<0.00l)* (0.122) (0.963)
2908 1587 3062 1691
3 Mean hours for two -0.021 -0.007 -0.058 -0.118way trip (cluster level) (0.033)* (0.572) (0.321) (0.193)
2633 1426 2773 1520
4 Density of cluster 0.001 0.002 -0.006 -0.003(cluster level) (0.280) (0.047)* (0.219) (0.731)
2750 1491 2892 1589
5 Price of wood (cluster -0.229 -0.175 -0.609 -0.017level) (<0.001)' (0.002)0 (0.017)* (0.963)
2139 1112 2234 1176
6 Problems with wood 0.120 0.100 0.093 0.158supply (t) (household (0.001)* (0.016)* (0.593) (0.503)level) 3030 1631 3184 1735
7 Hours for two way trip -0.015 -0.016 -0.012 -0.043(mean) (household (0.054) (0.089) (0.781) (0.511)level) 1624 899 1716 964
8 Kilometers to wood 0.001 0.003 0.017 -0.020source (household (0.882) (0.396) (0.430) (0.498)level) 1695 937 1792 964
Notes: (t) indicates variables for which parameter estimates reported in birth in the past 5 years model are for change in the variable from 0to 1. (*) indicates significant at the 95% level.
Source: PIHS
32
Table 12: Regionally disaggregated estimaes of the change in the probability of a birth in response to a change in ameasure of firewood scrcity in probit regressions of a Birth in the Past S Years on the sample of, for rual, ever marriedwomen, age 19 and over and the coefficients on measures of firewood scarcity when entered singly into OLS regressionsof hildren Ever Rorn for rural ever-marnied women 14 and over.
l_________ ___ Birth in the Past 5 Years l Children Ever Bom
Baluchi BaluchiPunjab Sindh NWFP stan Punjab Sindh NWFP stan
Variable added region region region region region region region region
1 Mean 0.023 1.100 -1.678 0.159 0.629 0.017 5.257 1.474problems (0.781) (<0.001)* (0.186) (0.576) (0.172) (0.987) (0.273) (0.333)with supply' 1477 846 488 219 1542 888 527 227
2 Mean -0.001 0.012 0.023 -0.0005 -0.031 0.037 0.031 0.139kilometers to (0.815) (<0.001)0 (0.002)' (0.976) (0.243) (0.112) (0.176) (0.124)soumce' 1407 843 460 198 1472 885 499 206
3 Mean hours -0.090 0.075 0.032 -0.057 -0.217 -0.024 0.021 0.067for two way (0.001)* (<0.001)* (0.209) (0.029)' (0.021)' (0.885) (0.831) (0.631)trip' 1281 755 424 173 1338 797 458 180
4 Density of -0.005 0.003 0.003 0.008 -0.012 -0.015 -0.005 0.017cluster' (0.001)* (0.141) (0.070) (0.074) (0.173) (0.470) (0.518) (0.478)
_____ _ 1436 607 488 219 1499 639 527 227
5 Price of -0.139 -0.359 -0.017 -0.217 -0.147 -0.256 -1.301 -2.237wood" (0.013)* (<0.001)* (0.923) (0.112) (0.661) (0.690) (0.035)' (0:001)*
_____ _ 1072 671 293 103 1116 700 309 109
6 Problems 0.076 0.148 0.206 0.130 0.439 -0.152 -0.094 -0.674with wood (0.106) (0.001)' (0.444) (0.317) (0.064) (0.619) (0.915) (0.311)supplyb (t) 1477 846 488 219 1542 888 527 227
7 Hours for -0.043 0.017 -0.016 -0.042 -0.152 0.179 -0.068 0.096two way trip| (0.001)' (0.251) (0.362) (0.156) (0.046)' (0.068) (0.330) (0.526)(mean) 727 506 283 108 762 535 306 113
8 Kilometers to -0.005 0.007 0.002 -0.013 -0.059 0.124 -0.137 0.026wood source' (0.344) (0.177) (0.880) (0.357) (0.084) (<0.001)' (0.008)* (0.746)
1 1l 764 1 533 1 287 1II 801 565 319 116
Notes: (t) indicates variables for which parameter estimates reported in birth in the past 5 years model are for change in thevariable from 0 to 1. (e) indicates significant at the 95% level. (a) Cluster level variable. (b) Household level variable.
Source: PIHS
33
Cnclusi'
This paper establishes several facts related to a vicious circle between environmental
degradation and fertility:
Firewood is an important part of the consumption bundle of rural households.
In many nrual regions firewood is often collected from open access or common or quasi-
common property. Poorer households are more likely to collect rather than purchase
firewood.
Children (at least the female children for which we have data) are relatively specialized
in collection activities, especially at young ages.
These facts enhance the plausibility of the general story that environmental scarcity could
possibly raise demand for children.
There are several findings however that may be inconsistent with a vicious cycle story:
* patterns of collecting versus purchasing firewood do not correspond in an obvious way
to demographic structure - households with more under 5 year olds and with more 11 to
15 year olds are significantly less likely to collect only, and those with more over 15 year
olds are more likely to collect only or to combine collecting and purchasing;
* while children's time allocation is relatively collection activity intensive, it is not
noticeably more so over the entire "child" period (say 9 to 18 years old) than women's
time, and women's time is also devoted to child rearing. A rise in the relative value of
collection activities would also therefore raise the price of time devoted to child activities;
* adding the available set of (admittedly very crude) proxies for the scarcity of firewood
to a simple regression does not yield results which are strikingly consistent with the
34
vicious circle hypothesis, although it does support the notion that in some regions the
hypothesis may hold.
This leads to three observations. First, there are a number of reasons to expect that we
would not be able to capture the impact of firewood scarcity on ferdlity in our data. We do not
have time series data on the scarcity in various regions and the scarcities may have changed
dramatically and differentially over the years. Therefore, we cannot relate the conditions on
firewood at the time to the fertility conditions at the time. In addition, the magnitude of the effect
itself is likely to be small. Even if firewood is a substantial fraction of the budget and even if
children are quite collection activity intensive, it is still the case that the change in the perceived
relative cost of children due to environmental changes is likely to be small. Searching for a
smallish effect with very noisy measures is likely to end in frustration.
Second, the failure to identify a large feedback effect from degradation to ferilit is no
cause for complacency. Even models with zero feedback or weak negative feedback could
produce unstable environmental conditions. Anything that reduces such mitigating factors
increases the likelihood of bad outcomes.
Third, a vicious circle between environmental degradation and the demand for children
will only exist in certain conditions which may change in part as an endogenous response to
population pressures (which are implied by increases in fertility). If the change is such that
households bear the costs of the environmental degradation directly (i.e. a switch from common
to private property regimes), then this will shift the effect on the demand for children. The
35
vicious circle will be as much the prodict of social and institutional a nts as an inrinsic
phenomenon.
36
Alderman, Harold and Salim Chishti, 1991, "Simultaneous deemination of household andmarket-oriented activities of women in rural Pakistan," in7: pp245-265.
Benefo, Kofi Darkwa and T. Paul Schultz, 1994, Determinants of fertility and child mortalityCote d'IyoiM and Ghana, LSMS Working Paper No. 103.
Cain, Michael, 1980, "The economic activities of children in a village in Bangladesh," in Biual1ousehold Studies in Asia, Singapore: Singapore University Press.
Cleaver, Kevin M. and Gotz A. Schreiber, 1994, Reversing the Spiral Te PpltioAgriculture and Environment Nexus in Sub-Saharan Africa (and Suplemnmm,Washington DC: World Bank.
Dasgupta, Partha, 1993, An Inquiry into Well-Being and Destitution, London: Oxford.
Dasgupta, Partha, 1995, "Population, Poverty, and the Local Environment,'" SontiAmcian272(2), pp 26-31.
Dasgupta, Partha and Karl-Goran Maler, 1995, "Poverty, Institutions, and the EnvironmentalResource Base,' in Handbook gf DevelopmentEconomics, Amsterdam: North Holland.
English, John, Mary Tiffen, and Michael Mortimore (in cooperation with ODI, London), 1994,"Land Resource Management in Machakosm District, Kenya, 1930-1990," World .BankE-nvironment Papr Number 5, Washington DC.
Grosh, Margaret and Paul Glewwe, 1995, A guide to Living Stan Masurement Stusurveys and their data sets, LSMS Working Paper No. 120.
Kumar, Shubh K. And David Hotchldss, 1988, "Consequences of deforestation for women's timeallocation, agricultural production, and nutrition in hill areas of Nepal," IFPRI ResearchReport #69, Washington DC.
Leach, Gerald, 1993, "Farm trees and wood markets: A review and economic appraisal, " mimeoin Pakistan Household Energy Strategy Study (HESS).
Montgomery, Mark and Aka Kouame, 1995, Fertility and child schooling in Cote d'lvoi=:khra tadeff, LSMS Working Paper No. 112.
37
Nag, Moni, Benjamin N.F. White and Creighton Poet, 1980, -An Anthropological aroach tothe study of the economic value of children in Java and Nepal", in }
nies in ia, Sngapore: Singapore University Press.
Nerlove, Marc, 1991, "Population and the Environment: A Parable of Firewood and OtherTales," Ameri=anJournal ofg tiurn 1nEonomics, 73: 133447.
Nerlove, Marc, 1993, "Procreation, Fishing, and Hundng: Renewable Resources and DynamicPlanar Systems," Amerian I of A tral1Rcnomics, 75: 59-71.
Ouerghi, Azedine, 1993, "Woodfuel use in Pakistan: Sustainibility of supply and socioeconomicand environmental implications," mimeo in Paki_an Household Enr Stategy Sudy
Ouerghi, Azedine and Charles Heaps, 1993, "Household energy demand: Consumption patterns"mimeo in Pak-istann Eoshold Eney Strategy Study UHSS.
Pstan Demograpic and Health Sue. 1990/91. IRD/Macro International.
Turner, B.L, Goran Hyden, and Robert W. Kates, 1993, 1=Jaion r=rowth anAggnal=Chang inAfrica, Gainesville: University Press of Florida.
World Bank, 1995, Pstan ngated Household SWe (PTHS) 1991a Basic Document, PolicyResearch Department, Poverty and Human Resources Division, mimeo.
38
ii KI I:I I I ~ I
I;c Ol. .- Lb j
I:t ~U) U) .
z
~ w U----- 9 a
Table AI.2: Housedold energy deanud, by activity and type of fiNd (in TOE)
Activy: Cool SpAce War Lihtn Space Oder Total
Woodfueis 8623 863 1064 0 0 206 10756(80.2%) (8.0%) (9.9%) (0.0%) (0.0%) (1.9%) (100%)[55.0%1 [75.2%] [71.4%] [0.0%] (0.0%] (31.1%] (53.8%]
Dung 3429 0 0 0 0 184 3613(94.9%) (0.0%) (0.0%) (0.0%) (0.0%) (5.1%) (100%)(21.9%] [0.0%] [0.0%] [0.0%] (0.0%] [27.8%] [18.1%]
Crop residues 2281 225 282 0 0 28 2816(81.0%) (8.0%) (10.0%) (0.0%) (0.0%) (1.0%) (100%)(14.5%] [19.6%] [18.9%] (0.0%] [0.0%] [4.2%] [14.1%]
Naural gas 1212 59 144 0 0 0 1415(85.7%) (4.2%) (10.2%) (0.0%) (0.0%) (0.0%) (100%)[7.7%] (5.1%] [9.7%] (0.0%] [0.0%] [0.0%] [7.1%]
Electicity 0 0 0 304 319 228 851(0.0%) (0.0%) (0.0%) (35.7%) (37.5%) (26.8%) (100%)
°(0.0%] 0.0%] [0.0%] [44.6%] [100%] t34.4%] [4.3%]
LPG 90 0 0 2 0 0 92(97.8%) (0.0%) (0.0%) (2.2%) (0.0%) (0.0%) (100%)(0.6%] [0.0%] [0.0%] (0.3%] (0.0%] [0.0%] (0.5%]
Kerosene 49 0 0 376 0 17 442(11.1%) (0.0%) (0.0%) (85.1%) (0.0%) (3.8%) (100%)
[0.3%] (0.0%] (0.0%] [55.1%] (0.0%] [2.6%] (2.2%]
Total 1586 1147 1490 682 319 663 19985(78.5%) (5.7%) (7.5%) (3.4%) (1.6%) (3.3%) (100%)f4OO%] (100%] (W0,Ir 100%] [100%] [100%] [100%]
Note: Figures in parenthesis are percent by activity, figures in brackes are percent by fuel
Source: Ouerghi and Heaps (1993)
40
Table A1.3: Total, moden, and biofuel use by urbn-rul resiec in Pakistn (In TOE)
Urbat Rual All Pakisan
Firewood 1708 8926 10634
Dung 511 3103 3614
Crop Residues 286 2529 2815
Charcoal 17 102 119
Sub-total biofuels (Percent) 222 (56%) 14660 (95%) 17182 (86.2%)
Sub-total modem fuels Perent) 1980 (44%) 765 (5%) 2745 (13.8%)
Total 4502 15425 19927
Source: Ouerghi and Heaps (1993)
Table Al.4: Mean and median firewood use per day per household in Paklistan, in housebolds usng firewood, by urban-rural residence and province.
Rural UrbanMean Median N Mean Median N
All Pakistan 7.4 6.0 2100 5.8 5.0 1051
Punjab 6.0 5.0 1022 5.2 4.9 553
Sindh 7.8 7.1 619 5.1 5.0 174
NWFP 9.1 6.4 321 5.7 4.5 203
Baluchistan 12.3 10.0 _ 138 9.3 10.0 121
Source: Author's calculations based on the PIHS
41
Table AL.S: Amog households dot use firewood, frction that collect, purca or bodi
iPurchaseonlyT Purchase and Collect only
AU Pakisawn _ _ _ _ _ _ _ _ _ _
Urban 1 79.9 T 4.31 159Rural 24.9j 10.9 64.2
Rural only, regionsl disaggregations |
Punisab rgon 27.7 12.7 59.6
Sindh _ _ _ _ _ ro 16.4 7.5 76.2
NWPP region 1 32.5 12.8 S4.7
flaluchistan region 23.4 8.5 68.1
Rural only, poomst 50th household expenditure per person percendle in each geographic locaion
Rural, a ___ __ r 23.6 9.6 66.9
Punjab region 29.6 11.2 59.2
Sindh region 13.6 4.7 81.7
NWFP region 26.5 14.5 39.0
Baluchistan region 16.9 9.9 73.2
Source: PIHS
42
AX=dix 2- Allocntion of e and female labor from
As mentioned above, the PIHS does not have infonnation on the time spent by boys in houehold
activities only labor activities. Table A2. 1 shows the results from other studies on the relative time spent
by males and females in household and directly productive activities. The time spent by males in household
activities ranges from one tenth of the time females spend on these activities in the Bangladeshi and (with
more variability by age) the Indonesian villages studied, to about half in the Nepalese village studied. If
these studies are representative, then males at all ages contribute only a small fraction of the time to these
types of activities and so our results may not be too badly biased by the limitation of the data.
The results on directly productive activities are more mixed. The average hours spent by males and
females is roughly on par in the Indonesian and Nepalese village studied, but they are as much as five times
as high in the Bangladeshi village. As seen in the main text, this disparity is similar to what occurs in rural
Pakistan
Of course, the allocation of time across activities depends on the definitions of household and non-
household activities used which vary somewhat across the different studies swmmarized in table A2. 1. In
the Cain (1980) study, "housework" includes housekeeping, food preparation, shopping, child care, and
"other useful and necessary, but not directly producdve activities", and "productive' includes animal care,
crop production, wage work, trading, fishing, other (handicraft production for consumption or sale, own
hut construction, self-employed sldlled work, and exchange labor). In the Nag, White and Peete (1980)
study "household maintenance" includes child care, household food preparation, firewood collection, and
other household maintenance work, and "directly productive" includes animal care, wage labor,
handicrafts, labor exchange, rice and garden cultivation, trading, preparation of food for sale, other.
43
Table A2. 1: Ratio of ave hours spau by males reladve to that of femles, by activity from ode saudi. in Asiancountri_ _ __ _ _
Age group Type of task: Total Number of
Housebold Non-Household (Male /Work Production Female)
Baruadeshi village 10 to 1.S. 0.1 4.7 1.1 165/ 189
16to21 0.1 5.0 1.0 108/ 77
22to59 0.1 4.4 1.0 4101441
Indonesian village 9 to 14 0.3 0.4 0.4 17110
15 to 19 0.1 1.2 0.8 6/5
20to29 0.1 1.2 0.7 5/9
30 and over 0.2 1.5 0.8 20 19
Nepalese village 9 to 14 0.5 0.7 0.6 82 55
15to 19 0.6 0.9 0.8 29/ 33
20to29 0.4 1.0 0.9 20/ 28
30 and over 0.4 1.1 0.8 56 / 110
Source: Rngladash is adapted from Cain (1980), Indonesia and Nepal from Nag, White and Peet (1980)
Note: See text for definitions of household and non-household work.
44
Alppendix 3: Latbior time use ofmalesAnd fmales
Figure A3.1 shows the age profiles of reported hours in labor activities (work on family farm, in
family enterpnse, farm wage work, non-fam wage work) for fmales and males above ten years old from
the PIHS data.
F3 A.1
Hours per week spent by females and males in labor activitin
so
34 -- i -Xv
XF30 - --------------- --------------- ------------------------------------------------------------------------- ----1 - le ;30 ----------
A25 - ---------------1-- ---------------------------------------------------------------------------------------------------|real2F - - - _
Age
Figure A3 is striking in the differental it shows betwveen males and females, even at very young
ages. Boys aged fifteen report working on average 15 hours per week in labor activities whereas girls
report approximately 2.5 hours per week. By age 25, the average number of hours women report in labor
activities reaches a plateau slightly above 5 hours per week. For males, the average nmmber of reported
hours per week increases sharply and cornsistently to about 35 hours at age 25, and peaks at about 45 hours
at age 35.
45
Table A3.1 presents the avcage hours per week spent by nual males and femia betwee the ages
of 10 and 25 in each of the four labor activity categories. Table A3.1 suggests that if females under 25
in rural areas are engaged in labor acdvities, it is on the family's own farm as the hours reported for
females in other labor activities is close to zero. For males the paerm is differe as mral males under 25
work initally in the family's own farm, but then work iwreasingly in family enterprise and for wages on
other farms. These results for females are consistent with what was found by Alderman and Chishti (1991)
in a study on the simultaneous determination of time in household and market oriented work amongst ural
women in Pakistan.
Table A3.l1: Average howr per week spewt by rural fermals and males in echCI lbor actvity
Females Males
Age Range: 10 to 14 1S to 19 20 to 24 10 to 14 IS to 19 20 to 24
Hours on own farm 2.76 3.92 4.36 5.01 9.72 9.93
Hours in familyenterprise 0.21 0.69 0.95 0.84 3.44 6.36
Hours in farm wagework 0.14 0.44 0.55 1.59 8.33 17.40
Hours io non-farmwage work 0.00 0.00 0.00 0.03 0.31 0.31
Total.15.05 5.86 7.47 21.8 34
Note: Hours in own farm and own enterprise are for pas 7 days, hours in frm and non-frm wage work are "usualhours.
Source: PIHS
46
Annidix 4: Reded form equamons for ferility
Table A4. I reports OLS esdmates of the determinants of children ever born (CEB) and of the
probability of observing a birth in the past 5 years (B5Y), in a reduced form&. The function is estimated
first for all women (with a dummy variable for rral residence) and then only for ural women. Table
A4.2 reports the results when the log of household expendiures per adult is excluded, and when the sample
is limited to those in the poorest expenditure group.
Pakistan is a high fertility country. The 1990/91 Pakistan Demographic and Health Survey
calculated a total ferdlity rate of 5.4 and mean nber of children ever born to women 40 to 49 of 6.55.
Ever married women in the PIHS sample have a total ferdlity rate of 6.0 in urban areas and 5.5 in mral
areas. The mean number of children born to women 40 to 49 is 6.9 in urban areas and 7.1 in ural areas.
Larger total household expenditures per adult are associated with higher CEB when all income
groups are pooled. The result is insignificant when only the poorest 50 percentile in the rural areas are
considered. Expenditures are associated with a higher probability of B5Y for those women in households
which are in the poorest 50th percentile in rural areas. This is consistent with a positive income elasticity
for children, especially after controlling for education effects which capture to some extent the value of
time of mother's and father's time.
However, since children may contribute to family income, the causal direction is not clear2s. The
findings in the household fertility literature on income have been mixed26. When estimated using two-stage
least squares (with the head of household's characteristics, housing characteristics, and asset ownership
varables as instmments), expenditures have an insignificantly negative effect on CEB for women in rural
I This is called a demand function mostly out of an empirically informed prior that differences indemand account for by far the largest fraction of fertility variation (Pritchett, 1994).
2' Montgomery and Kouame (1994) have a useful discussion of this causal effect. They explore theexogeneity of expenditures per adult using instrumental variables (IV) in a very similar regression usingdata from Cote d'Ivoire. They conclude that '...it is disturbingly vulnerable to the selection of instrumentsand to multicollinearity. In many applications, it is not clear that the method [N brings one any closerto the truth'. Appendix table I shows the results excluding the expenditures variable
26 Montgomery and Kouame (1994) find a positive and significant effect in Cote d'Ivoire. Benefo andSchultz (1994) use nearly identical regressions and yet find a positive and signif;icant effect in Cote d'Ivoireand a negative and insignificant effect in Ghana. After instrumentng to account for the endogeneity ofincome the negative effect in Ghana becomes significantly different from zero.
47
areas". When estimatng B5Y and log of expenditures per adult jointly (using the same set of insuments
as CEB), expenditures have a negative and insignificant effect on B5Y for women in rural areas.
However, a test of the correlation between the errors from the two equations rejects endogeneity.
Estirmtes on the non-income variables are not greatly affected by treating expenditures as endogenous.
The age effects reflect the child bearing cycle: it's quadratic nature captures the fact that CEB,
as well as the likelihood of having a birth, increases up to a certain age and then levels off.
Education for the woman is limited to the "ever went to school" variable. This is because there are too
few observations on non-zero years of schooling to reasonably identify their effect. For example, among
the ever-married rural women aged 14 and over in the CEB regression, 92.4 percent have zero years of
schooling, and 98.1 percent have 6 or fewer years of schooling. As is common in the literaure, the results
show strong negative effects of female education on the number of children ever born. Having gone to
school reduces the number of children ever born by a little more than half a child on average. The effect
on the probability of birth in the past 5 years is negative, although the parameter estimates are not
significantly different from zero. All of the husband's education parameter estimates are insignificantly
different from zero in the rural samples.
The Punjab province is the excluded provincial dummy variable, hence the results indicate that
women in the Sindh region have slightly lower fertility, and those in the NWFP have slightly higher
fertility than in the Punjab region. Surprisingly, rurl women have fewer children, after correcting for the
other factors.
27 The two-stage least squares and joint-estimation results are not reported here but are available fromthe authors.
48
Table A4.1: OLS esdmats of the detrminan of children ever bom for evr married women aged 14 and over, and wean change inIthe probability of birth in the past S ar (from Probkt estimation) for ever married wmnaged 19 and Over.I
Birth in the Past 5 Years Children Ever Born
Rural and Urban Rural only Rural and Urban Rural only
Household expenditures per adult, 0.021 -0.002 0.158 0.252natual log (0.062) (0.889) (0.003)' (0.001)'
Age in years 0.084 0.088 0.555 0.570(<0.001)' (<0.001)* (<0.001)' (<0.001)*
Age in years. squared -0.002 -0.002 -0.005 -0.005(<0.001)' (<0.001)' (<0.001)* (<0.001)*
I -Ever went to school (t) -0.008 -0.026 -0.831 -0.519(0.678) (0.402) (<0.001)# (0.006)#
1-Spouse information available -0.295 -0.211 0.520 0.116(t) (0.244) (0.405) (0.572) (0.907)
Spouse years of schooling 0.004 -0.001 0.070 0.022(0.350) (0.877) (0.002)* (0-544)
Spouse years of schooling, squared -0.001 -0.0003 -0.010 -0.004(0.028)- (0.603) (<0.001)' (0.195)
I -Sindh region (t) -0.203 -0.252 0.161 0.157(<0.001)' (<0.001)* (0.047)* (0.165)
1 -NWFP region (t) 0.068 0.025 0.372 0.238(0.001)* (0.325) (<0.001)' (0.082)
1 -Baluchistan region (t) -0.063 -0.069 0.116 -0.131(0.006*) (0.027)' (0.367) (0.491)
I-Rural community (t) -0.031 -0.132(0.025)_ (0.064)
Constant -9.007 -9.622(<0.001)' (<0.001)'
Summary statistics
N 5833 3030 6085 3184
R-Squared (pseudo R-Squared forBSY) 0.3132 0.3096 0.4401 0.4405
Mean of dependent variable 0.456 0.439 5.03 5.02
Notes: (t) indicates variables for which parameter estimates reported in birth in the past 5 years model are for change in the variablefrom 0 to 1. P-values (of probit parameter estimates) are reported in parenthesis, C') indicates significanly different from zero at the95% level
Source: PIHS
49
Table A4.2: OLS edmaes of the deminant of children ever born for ever married women aged 14 and over, and mean du inthe probability of birth in the past 5 yeas (from Probit esimadon) fhr evr marrid ne aged 19 od over.
Birth in the Past 5 Years Childreo Ever Born
Rural only Rural only
Poorest 50th Poores 50thAll percentile (1) All perceiloe ()
Household expeditures per adult, 0.074 0,075naural log (0.014)* (0.676)
Age in years 0.088 0.074 0.580 0.548(<0.001)* (<0.001)* (<0.001)* (<cO.0l)*
Age in years, squared -0.002 -0.001 -0.005 O-O.O5(<0.001)* (<0.001)* (<o.o00)* (< 0.001)
1 -Ever went to school (t) -0.027 -0.043 -0.457 -0.627(0.390) (0.303) (0.016)* (0.038)'
-Spouse infomation available -0.212 -0.194 0.165 -0.298(t) (0.403) (0.430) (0.869) (0.804)
Spouse years of scowoling -0.001 0.0001 0.264 0.008________=________________ (0.875) (0.994) (0.468) (0.881)
Spouse years of schooling, squared -0.0003 -0.001 -0.004 -0.002|_______________ (0.602) (0.499) (0.205) (0.644)
I -Sindh region (t) -0.252 -0.249 0.173 0.089(<0.001)* (<0.001)* (0.125) (0.553)
I-NWFP region (t) 0.025 -0.002 0.318 0.123(0.329) (0.943) (0.018)* (0.544)
! -Baluchistan region () -0.070 0.004 -0.072 40.140(0.026)* (0.928) (0.707) (0.629)
1-Rural community (t)
Constant -8.224 -7.821(<0.001)* (<0.001)'
Summary statistics
N 3030 1631 3184 1735
R-Squared (pseudo R-Squared forBSY) 0.3096 0.3536 0.4386 0.4719
Mean of dependent variable 0.439 0.405 5.02 4.81
Notes: (#) Poorest 50th percentile as measured by household expenditures per adult (t) indicates variables for which parmmeterestimates reported in birth in the past 5 years model are for change in the variable from 0 to 1. P-values (of probit parameterestimates) are reported in parenthesis, (7) indicates significantly different from zero at the 95% level.
Source: PIHS
50
Policy Research Working Paper Series
ContactTitle Author Date for paper
WPS1599 Economic Implications for Turkey Glenn W Harrison May 1996 M Patenaof a Customs Union with the Thomas F Rutherford 39515European Union David G Tarr
WPS1600 Is Commodity-Dependence Nanae Yabuki May 1996 G llogonPessimism Justified? Critical Takamasa Akiyama 33732Factors and Government Policiesthat Characterize Dynamic CommoditySectors
WPS1601 The Domestic Benefits of Tropical Kenneth M Chomitz May 1996 PRDEIForests: A Critical Review Kanta Kumari 80513Emphasizing Hydrological Functions
WPS1602 Program-Based Pollution Control Shakeb Afsah May 1996 A WilliamsManagement The Indonesian Benoit Laplante 3/176PROKASIH Program Nabiel Makarim
WPS1603 Infrastructure Bottlenecks, Private Alex Anas May 1996 S WardProvision, and Industrial Productivity: Kyu Sik Lee 31707A Study of Indonesian and Thai Michael MurrayCities
WPS1604 Costs of Infrastructure Deficiencies Kyu Sik Lee May 1996 S `Wardin Manufacturing in Indonesia, Alex Anas 31707Nigeria, and Thailand Gi-Taik Oh
WPS1605 Why Manufacturing Firms Produce Kyu Sik Lee May 1996 S WardSome Electricity Internally Alex Anas 31707
Satyendra VermaMichael Murray
WPS1606 The Benefits of Alternative Power Alex Anas May 1996 S WardTariffs for Nigeria and Indonesia Kyu Sik Lee 31707
WPS1607 Population Aging and Pension F. Desmond McCarthy May 1996 M DivinoSystems Reform Options for China Kangbin Zheng 33739
WPS1608 Fiscal Decentralization Public Tao Zhang May 1996 C BernardoSpending, and Economic Growth Heng-fu Zou 37699in China
WPS1 609 The Public Sector in the Caribbean: Vinaya Swaroop May 1996 C BernardoIssues and Reform Options 31148
WPS1610 Foreign Aid's Impact on Public Tarhan Feyzioglu May 1996 C BernardoSpending Vinaya Swaroop 31148
Min Zhu
Policy Research Working Paper Series
ContactTitle Author Date for paper
WPS1611 Economic Analysis for Health Jeffrey S. Hammer May 1996 C. BernardoProjects 37699
WPS1612 Stock Market and Investment: The Cherian Samuel May 1996 C. SamuelSignaling Role of the Market 30802
WPS1613 Does Public Capital Crown Out Luis Serven May 1996 E. KhinePrivate Capital? Evidence from India 37471
WPS1614 Growth, Globalization, and Gains Thomas W. Hertel May 1996 A. Kitson-Waltersfrom the Uruguay Round Christian F. Bach 323947
Betina DimarananWill Martin
WPS1615 Issues in Measuring and Modeling Martin Ravallion June 1996 P. SaderPoverty 33902
WPS1616 Transient Poverty in Rural China Jyotsna Jalan June 1996 P. SaderMartin Ravallion 33902
WPS1617 Why is Unemployment Low in the Simon Commander June 1996 L. AlsegafFormer Soviet Union? Enterprise Andrei Tolstopiatenko 36442Restructuring and the Structureof Compensation
WPS1618 Analytical Aspects of the Debt Stijn Claessens June 1996 R. VelasquezProblems of Heavily Indebted Enrica Detragiache 39290Poor Countries Ravi Kanbur
Peter Wickham
WPS1619 Capital Flows to Latin America: Sara Calvo June 1996 M. GomezIs There Evidence of Contagion Carmen Reinhart 38451Effects?
WPS1620 Bank Insolvencies: Cross-country Gerard Caprio, Jr. July 1996 B. MooreExperience 38526
WPS1621 The Sustainability of African Debt Daniel Cohen July 1996 S. King-Watson33730
WPS1622 Capital Control Liberalization and Ross Levine July 196 P. Sintim-AboagyeStock Market Development Sara Zervos 38526
WPS1623 Environmental Degradation and Deon Filmer July 1996 S. Fallonthe Demand for Children: Searching Lant Pritchett 38009for the Vicious Circle